| Sumario: | Part of a special issue on computer-based performance assessment of problem solving. The writers examine the ability of artificial neutral network technologies to generate performance models of complex problem-solving tasks without detailed a priori knowledge of the nature of the task. They then apply this analysis to two different content domains—clinical patient management and high school genetics—in order to test the generalizability of the approach. Their analysis indicates that in both domains, the artificial neural networks, which used only the sequence of actions taken while performing the task, generated multiple classification groups defining different levels of competence. They point out that the good concordance of these classifications with independently derived expert ratings further establishes the validity of these neutral network performance groupings.
|
|---|